Branched Phospholipids Render Lipid Vesicles More Susceptible to Membrane-active Peptides

Abstract

Iso- and anteiso-branched lipids are abundant in the cytoplasmic membranes of bacteria. Their function is assumed to be similar to that of unsaturated lipids in other organisms – to maintain the membrane in a fluid state. However, the presence of terminally branched membrane lipids is likely to impact other membrane properties as well. For instance, lipid acyl chain structure has been shown to influence the activity of antimicrobial peptides. Moreover, the development of resistance to antimicrobial agents in Staphylococcus aureus is accompanied by a shift in the fatty acid composition towards a higher fraction of anteiso-branched lipids. Little is known about how branched lipids and the location of the branch point affect the activity of membrane-active peptides. We hypothesized that bilayers containing lipids with low phase transition temperatures would tend to exclude peptides and be less susceptible to peptide-induced perturbation than those made from higher temperature melting lipids. To test this hypothesis, we synthesized a series of asymmetric phospholipids that only differ in the type of fatty acid esterified at the sn-2 position of the lipid glycerol backbone. We tested the influence of acyl chain structure on peptide activity by measuring the kinetics of release from dye-encapsulated lipid vesicles made from these synthetic lipids. The results were compared to those obtained using vesicles made from S. aureus and S. sciuri membrane lipids extracts. Anteiso-branched phospholipids, which melt at very low temperatures, produced lipid vesicles that were only slightly less susceptible to peptide-induced dye release than those made from the iso-branched isomer. However, liposomes made from bacterial phospholipid extracts were generally much more resistant to peptide-induced perturbation than those made from any of the synthetic lipids. The results suggest that the increase in the fraction of anteiso-branched fatty acids in antibiotic-resistant strains of S. aureus is unlikely to be the sole factor responsible for the observed increased antibiotic resistance.

INTRODUCTION

Membrane-active peptides are ubiquitous in all living organisms where they fulfill a variety of functions. They contribute to intercellular communication, are secreted for offensive and defensive purposes by bacteria, plants, and fungi, and are part of the innate immune systems in animals (1–3). Peptides with antimicrobial properties preferentially interact with the cytoplasmic membranes of bacteria and other microorganisms, and, thus, provide a blueprint for the development of novel antibiotics – at least in principle. The elucidation of the mode of action of specifically antimicrobial peptides has been the subject of intense research for about three decades. However, progress in the synthesis and use of antimicrobial peptides as antibiotics has been slow and our understanding of the factors that govern their target specificity remains incomplete. The issue is further complicated by an often observed lack of correlation between results from model studies and the in vivo activities of antimicrobial peptides.

The interaction between antimicrobial peptides and their target membranes is most often studied using mixtures of unsaturated phospholipids, with headgroup compositions that reflect those of the major lipid classes found in bacterial cytoplasmic membranes. However, bacterial lipid species are more diverse and bacterial cytoplasmic membranes more complex than suggested by these simple model systems. The majority of phospholipids that comprise the single cytoplasmic membrane of Staphylococcus aureus are diacylphosphatidylglycerols (PG) or derived from PG (4). Among these are cardiolipin and lysylphosphatidylglycerol (LPG), where the PG headgroup is esterified to the carboxyterminus of a lysine residue. Monounsaturated palmitoyl and oleoyl chains exist in only negligible amounts in the genus Staphylococcus and are altogether absent from S. aureus (4). In S. aureus, membrane lipids are entirely saturated, with terminally methyl-branched fatty acids functionally replacing the monounsaturated chains found in other organisms (4–6). The most abundant branched fatty acids found in S. aureus are anteiso (ai) and iso-branched (i), C15 and C17 chains. In S. epidermidis, Bacillus spp., and Streptococcus spp. longer chain fatty acids (≥ C17), normal or iso-branched, are found preferentially in the sn-1 position (7–10) and preliminary data suggests that this is also true for S. aureus. Shorter (C15) iso- and anteiso-branched chains are more commonly found in the sn-2 position.

The degree of branching and the position of the branch point influences the main phase transition temperature (T_m) of hydrated phospholipid vesicles (11–16). Symmetric di-anteiso-branched phosphatidylcholines (PC) with C16 hydrocarbon chains or shorter show a main T_m below 0°C, whereas di-ai17PC melts around 10°C. The transition enthalpies of ai-branched phospholipids are significantly smaller than for their straight-chain equivalents, indicating that ai-branched PC form a more disordered gel-phase than straight-chain PCs. By comparison, di-iso-branched PCs melt at higher temperatures than their anteiso-branched counterparts and their transition enthalpies are comparable to those of straight-chain PCs.

The susceptibility of S. aureus to antimicrobial peptides correlates with the fraction of anteiso-branched in the bacterial cytoplasmic membrane (17–20). In methicillin-resistant S. aureus (MRSA), the fraction of anteiso-branched acyl chains is elevated relative to that of susceptible strains and growth of S. aureus in medium containing sublethal concentrations of methicillin or antimicrobial peptides induces an increase in the anteiso-branched fatty acid fraction. Moreover, decreasing the anteiso-branched fatty acid content in S. aureus by the inactivation of a key enzyme in the synthetic pathway has been shown to lead to reduced virulence (21), suggesting a causal link between acyl chain structure and bacterial resistance to antibiotics.

We showed previously that the action of the cytolytic peptide δ-lysin strongly depends on the acyl chain structure of the target membrane in lipid vesicles (22). Specifically, we found that membrane perturbation is inhibited in proportion to the degree of fatty acid unsaturation. We attributed this observation to an entropically unfavorable interaction of the peptide with highly disordered, unsaturated acyl chains. These results suggested that a shift in the lipid acyl chain composition toward a higher fraction of acyl chains with a low T_m should increase bacterial resistance to antimicrobial peptides. Based on these observations, we hypothesized that δ-lysin would also interact less favorably with lipid vesicles composed of anteiso-branched acyl chains than with those composed of straight-chain and iso-branched phospholipids of similar chain length due to the lower T_m of the anteiso-branched variants.

The hypothesis was tested by synthesizing asymmetric saturated PCs that differed only in the type of fatty acid in the sn-2 position: 1-palmitoyl-2-(12-methylmyristoyl)-sn-glycero-3-phosphocholine (16:0-ai15:0PC), 1-palmitoyl-2-(13-methylmyristoyl)-sn-glycero-3-phosphocholine (16:0-i15:0PC), and 1-palmitoyl-2-tridecanoyl-sn-glycero-3-phosphocholine (16:0-13:0PC) (Fig. 1).

Figure 1.

Structures of synthetic lipids. (A) 1-palmitoyl-2-(12-methylmyristoyl)-sn-glycero-3-phosphocholine (16:0-ai15:0PC). (B) 1-palmitoyl-2-(13-methylmyristoyl)-sn-glycero-3-phosphocholine (16:0-i15:0PC). (C) 1-palmitoyl-2-tridecanoyl-sn-glycero-3-phosphocholine (16:0-13:0PC).

To test peptide–lipid interactions as a function of acyl chain branching, we encapsulated the fluorescent dye carboxyfluorescein in lipid vesicles and measured the kinetics of dye efflux induced by δ-lysin. The results were compared with those obtained using 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC). To assess if the synthetic lipids capture the behavior of bacterial lipids, we performed the same experiments using phosphatidylglycerol extracts from S. aureus and S. sciuri cell cultures and compared the peptide-induced dye efflux kinetics with those collected using 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-(1’-rac-glycerol) (POPG) vesicles. We observed that dye release from lipid vesicles composed of asymmetric, iso-branched phospholipids was generally slower than that from anteiso-branched lipid vesicles; however, the overall effect was minor. Moreover, liposomes composed of synthetic asymmetric branched chain phospholipids were significantly more prone to release content than POPC liposomes or those made from bacterial phospholipid extracts.

METHODS

Chemicals

1-Palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC), 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-(1’-rac-glycerol) (POPG) and 1-palmitoyl-2-hydroxy-sn-glycero-3-phosphocholine (C16-lyso-PC) were purchased from Avanti Polar Lipids (Alabaster, AL). Carboxyfluorescein (99% pure, lot A015252901) was purchased from ACROS (Morris Plains, NJ). Tridecanoic acid was purchased from MP Biomedicals (Solon, OH), 13-methylmyristic acid, 12-methylmyristic acid, methyl 12-methylmyristate, methyl 13-methylmyristate, and boron trifluoride-methanol (14% solution) were purchased from Sigma-Aldrich (St. Louis, MO). Lipids and fatty acids were tested by TLC and used without further purification. 4-pyrrolidinopyridine (98% pure) was purchased from TCI America (Portland, OR) and purified by recrystallization in hexane prior to use. The purified crystals were stored under argon. 2-(1H-Benzotriazole-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate (HBTU) was purchased from Novabiochem (Darmstadt, Germany), analytical thin layer chromatography (TLC) silica plates from Analtech (Newark, DE), flash grade silica (32–63 μ) from Dynamic Adsorbents (Norcross, GA), Molecular sieves 3A from Sigma-Aldrich (St. Louis, MO), and flash grade basic alumina (adsorption 80–200 mesh) from Fisher Scientific (Pittsburgh, PA). Chlorinated solvents and methanol (High performance Liquid Chromatography/American Chemical Society grade) were purchased from Burdick & Jackson (Muskegon, MI). Hexanes and trifluoroacetic acid (99% pure) were purchased from Mallinckrodt Chemicals (Phillipsburg, NJ). BBL Mueller Hinton (MH) Broth was purchased from Becton, Dickinson and Comapny (Sparks, MD), and celite 545 from Fisher Scientific (Fair Lawn, NJ). Bacterial cultures were from ATCC (American Type Culture Collection, Manassas, VA). δ-Lysin was a gift from Dr. Birkbeck, University of Glasgow. Stock solutions of MasX and CecA were prepared by dissolving lyophilized peptide in deionized water/ ethanol, 1:1 (v/v) (AAPER Alcohol and Chemical Co., Shelbyville, KY). Stock solutions of δ-lysin were prepared in deionized water, pH ~ 3. Peptide stock solutions were stored at −20 C and kept on ice during experiments. Peptide concentration of the stock solutions were determined precisely by measuring the absorbance at 280nm, and using a molar extinction coefficient of tryptophan of 5600 M⁻¹cm⁻¹.

Lipid synthesis

Asymmetric lipids were synthesized by coupling the desired fatty acid to the sn-2 position of C16-lyso-PC (manuscript in preparation). Briefly, CH₃Cl solutions of fatty acid and HBTU were combined in a 1:0.8 molar ratio in a dry round bottom flask. C16-lyso PC (in a molar ratio of 0.3:1 lysoPC to fatty acid), 4-pyrrolidinopyridine and TFA were combined in a second flask and slowly added to the fatty acid/HBTU mixture. The reaction mixture was heated to 40°C and monitored by TLC on alumina plates in CH₂Cl₂/methanol/water 80:20:2. Reaction yield was judged to be ≥ 90% by TLC. The reaction was stopped by the addition of a 1:1 mixture of CH₃Cl/methanol and concentrated by rotary evaporation. The sample was dried on a Welch Duo-Seal high vacuum pump (Niles, IL), dissolved in a minimal amount of CH₂Cl₂/methanol 9:1, and purified by flash column chromatography (FCC) on a basic alumina column.

Isolation of staphylococcal membrane lipids

Bacterial cultures were obtained from the American Type Culture Collection (ATCC, Manassas, VA). Staphylococcus aureus subsp. aureus Rosenbach ATCC 12600 or Staphylococcus sciuri subsp. sciuri ATCC 29062 were cultured in 2 L MH broth for 24 hours at 37 °C. Cells were harvested by centrifugation and resuspended in methanol. CH₂Cl₂ was added to the suspension to a final ratio of 1:2 CH₂Cl₂:methanol. The mixture was sonicated for 5 min. in a water bath sonicator and stored for 12–24 hours at 4 °C. The suspension was filtered into a round bottom flask through a celite pad in a 60 mL, medium-porosity Büchner funnel. The filtrate was reduced to dryness in a rotary evaporator, dried under high vacuum, and redissolved in approximately 2 mL 2:1 CH₂Cl₂/methanol. This crude lipid extract was separated into its constituent lipid classes by FCC on silica using a CH₂Cl₂/methanol/water gradient. Fractions were analyzed by TLC and phospholipids visualized with a modified Dittmer-Lester reagent (23, 24). Pure phosphatidylglycerol fractions were combined, reduced to dryness in a rotary evaporator and stored in CHCl₃ at −20°C. Lipid concentrations were determined by the Bartlett phosphate method (25), modified as previously described (26).

Fatty acid analysis of bacterial membranes

Approximately 20 mg of a bacterial lipid extract was saponified in 1 mL 0.5 M NaOH-methanol at 70–100°C. The free fatty acids were converted to their methyl esters (FAME) by refluxing in 1.5 mL BF₃-methanol for 30 minutes at 70–100°C (27). FAME were extracted in hexane and passed through a mini column made from a pasteur pipet, packed with 32–63 µ silica gel, to remove impurities, and concentrated for analysis by GC-FID. FAME were analyzed on a Hewlett Packard 6890 series gas chromatograph (Agilent Technologies, Santa Clara, CA) using an Agilent 19091J-413 HP-5, 30 m column. The samples were applied to the column using a split ratio of 10:1 and a flow rate of 1.0 mL/min. The initial GC oven temperature of 170°C was held for 2 minutes, ramped at 3°C/min to 200°C, and then increased at 15°C/min to a final temperature of 260°C, which was maintained for 2 minutes. Fatty acids were identified by the comparison of retention times of peaks in the sample with those of commercially available standards. Percent fatty acid values were calculated by integrating peak areas in the chromatogram.

NMR spectroscopy

Successful addition of a fatty acid at the sn-2 position of C16-Lyso-PC was determined by ¹H NMR using a 400Mhz or 600Mhz Bruker Spectrospin NMR spectrometer (Bruker Corporation, Billerica, MA). NMR samples were dissolved in 10% deuterated methanol in CDCl₃. The signal in the region of δ_H 2.2–2.4 ppm was assigned to the methylene protons located on the carbon α to the fatty acid carbonyl group. The signal change from a triplet in C16-Lyso-PC to a doublet of triplets in the diacyl-PC is indicative of the successful addition of a fatty acid at the sn-2 position. ³¹P NMR was performed to ensure that no residual C16-Lyso-PC was present in the sample.

Preparation of large unilamellar vesicles

Large unilamellar vesicles (LUVs) were prepared by mixing the lipids in chloroform in a round-bottom flask. For vesicles containing 7MC-POPE, the probes were added to the lipid in chloroform solution at a final probe concentration of 2 mol%. The solvent was rapidly evaporated using a rotary evaporator (Büchi R-3000, Flawil, Switzerland) at 60°C. The lipid film was then placed under vacuum for 4 hours and hydrated by the addition of buffer containing 20 mM MOPS, pH 7.5, 0.1 mM EGTA, 0.02% NaN₃, and 100 mM KCl or appropriately modified as indicated below. The suspension of multilamellar vesicles was subjected to five freeze-thaw cycles. The suspension was then extruded 10 × through two stacked polycarbonate filters of 0.1 µm pore size (Nuclepore, Whatman, Florham, NJ), using a water-jacketed high pressure extruder (Lipex Biomembranes, Inc., Vancouver, Canada) at room temperature. Lipid concentrations were assayed as previously described (26).

Peptide-induced dye release from lipid vesicles

LUVs for carboxyfluorescein (CF) efflux kinetics measurements were prepared by hydrating the dried lipid film with CF-containing buffer (20 mM MOPS pH 7.5, 0.1 mM EGTA, and 0.02% NaN₃, 50 mM CF) to give a final lipid concentration of 5 mM. Following extrusion, CF-containing LUVs were passed through a Sephadex-G25 column to separate the dye in the external buffer from the vesicles. The suspension was diluted in buffer to the desired lipid concentration and used for fluorescence measurements. The buffer used was 20 mM MOPS pH 7.5, containing 100 mM KCl, 0.1 mM EGTA, and 0.02% NaN₃, which has the same osmolarity as the CF-containing buffer. The kinetics of carboxyfluorescein efflux were recorded in an Applied Photophysics SX.18MV stopped-flow fluorimeter. CF was excited at 470 nm and the emission recorded through a GG 530 long-pass filter (Edmund Industrial Optics).

Calculation of average relaxation times

The curves of carboxyfluorescein release as a function of time were characterized by a mean relaxation time (τ), as described before (28). Briefly, the mean relaxation time is obtained from the integral (29, 30),

$$\tau =\frac{{\int }_0^{\infty }tf(t)dt}{{\int }_0^{\infty }f(t)dt},$$ (1)

where

$$f(t)=\frac{dF(t)}{dt}$$ (2)

and F(t) is the experimental curve of normalized fluorescence increase as a function of time. This curve increases as CF is released, until it essentially reaches a plateau (see Fig. 3). The time-derivative of F(t), f(t) behaves as the probability density function (29, 30). For example, for a multi-exponential decay τ is the weighted average of the relaxation times of each exponential function. If necessary, the curves were smoothened prior to numerical differentiation as described before (28), to avoid errors due to experimental noise.

Figure 3.

Carboxyfluorescein efflux from lipid vesicles induced by the cytolytic peptide δ-lysin as a function of lipid composition. 16:0-i15:0PC, blue trace; 16:0-13:0:0PC, green trace; 16:0-ai15:0PC, red trace; POPC, black trace. Peptide concentration was 0.5 µM and lipid concentration, 200 µM.

RESULTS & DISCUSSION

We used the cytolytic peptide δ-lysin, a 26-residue peptide that forms an amphipathic α-helix when bound to bilayer membranes, to probe the effect of lipid acyl chain structure on peptide activity. δ-Lysin is one of many toxins secreted by S. aureus and extremely efficient in permeabilizing the cell membranes of a wide variety of organisms and lipid vesicles with binding constants between 10⁵ and 10⁶ M⁻¹ (31–33). We showed previously that the mean time constant, τ, of dye efflux (Eq. 1) provides a model-independent and robust observable that is directly related to changes in bulk bilayer properties (22). Here, we postulated that the kinetics of peptide-induced dye release from lipid vesicles should depend on the type of acyl chain branching of the lipids comprising the vesicles. Specifically, we proposed that peptide–lipid interactions would be entropically unfavorable in systems composed of lipids with low phase transition temperatures. Thus, when comparing a series of lipids, dye release should occur slowest in vesicles made from lipids with the lowest T_m, if the hypothesis is correct.

POPC undergoes a gel-to-liquid phase transition at −3.5±1.0°C (34). We show here that aqueous lipid dispersions of 16:0-i15:0PC and 16:0-13:0PC undergo a cooperative phase transition with a T_m of 15.0°C for 16:0-i15:0PC and 18.4°C for 16:0-13:0PC (Fig. 2).

Figure 2.

Excess heat capacity curves of 16:0-i15:0PC and 16:0-13:0PC LUVs. (A) 16:0-i15:0PC, T_m = 15.0 ± 1.0°C, (B) 16:0-13:0PC, T_m = 18.4 ± 1.0°C. Both curves are the averages of two heating scans collected at a rate of 0.2°C/min, separated by a cooling scan at 0.2°C/min.

The main phase transition of the anteiso-branched variant 16:0-ai15:0PC occurs below 0°C (data not shown). The transition profile of 16:0-i15:0PC shows a single phase transition indicative of a well packed gel phase with a transition enthalpy ΔH, of 6.0 kcal/mol, comparable in magnitude to that of POPC (34). The overall width of the phase transition for 16:0-13:0PC is slightly larger than that for 16:0-i15:0PC and the corresponding ΔH is 3.7 kcal/mol. By comparison to the transition profile of 16:0-i15:0PC, the endotherm associated with the melting transition in 16:0-13:0PC vesicles appears to be a composite of several separate transitions, which is compatible with a partially interdigitated bilayer and multiple lipid packing arrangement in the gel phase (34, 35). Whole S. aureus membranes have been shown to exhibit a broad phase transition that is complete just below the growth temperature of the cell culture (36).

We observed that in the absence of peptide, LUVs composed of 16:0-13:0PC released encapsulated dye within tens of minutes (data not shown). To be able to measure peptide-induced dye release, 16:0-13:0PC LUVs were assayed immediately after the vesicles had been separated from the external dye solution. Nevertheless, signal strength was small and the experimental noise high, because a large fraction of the encapsulated dye had leaked from the vesicle lumen by the time peptide was added to the suspension (Fig. 3, green trace).

Encapsulated dye also leaked from 16:0-i15:0PC vesicles, but the process occurred on the order of hours rather than minutes, as it did for 16:0-13:0PC LUVs. Following the addition of τ-lysin, the suspension of 16:0-i15:0PC LUVs released dye rapidly (τ < 1s, Fig. 3, blue trace). δ-Lysin also acted very rapidly on 16:0-ai15:0PC vesicles, releasing content with a τ ≈ 2s, under the same conditions (Fig. 3, red trace). It is an observation of note that of the three lipids synthesized here, 16:0-ai15:0PC produced lipid vesicles that were much more tightly sealed than those composed of 16:0-13:0PC or 16:0-i15:0PC. Dye encapsulated 16:0-ai15:0PC LUVs retained dye for at least a day; yet, all three lipid species produce LUVs that are significantly more permeable to encapsulated dye than those composed of POPC, which retain dye for weeks. Peptide-induced dye release from of POPC vesicles, though still rapid, occurs with an average τ ≈ 7s (Fig. 3, black trace), which is an order of magnitude larger than that measured from 16:0-i15:0PC LUVs at a comparable lipid concentration.

A closer look at the kinetics of dye release as a function of lipid concentration shows that for LUVs composed 16:0-13:0PC and 16:0-i15:0PC, dye release occurs with kinetics that depend little on lipid concentration (Fig. 4). Clearly, these lipids produce vesicles that are intrinsically permeable to carboxyfluorescein and very susceptible to structural perturbations, including those induced by membrane-active peptides. The anteiso-branched variant 16:0-ai15:0PC, on the other hand, produces LUVs that qualitatively behave like those based on POPC with respect to peptide-induced dye release in that the kinetics of dye release become slower as the lipid concentration increases (Fig. 4). Nevertheless, at high lipid concentrations, dye release occurs significantly faster from 16:0-ai15:0PC vesicles than from POPC vesicles at the same lipid concentration (Fig. 4). This observation suggests that formation of the state responsible for dye release, most likely a small peptide oligomer (26), is favored in 16:0-ai15:0PC over POPC.

Figure 4.

Average kinetics (τ) of peptide-induced carboxyfluorescein efflux from lipid vesicles as a function of lipid concentration. 16:0-i15:0PC, blue symbols; 16:0-13:0:0PC, green symbols; 16:0-ai15:0PC, red symbols; POPC, black symbols. Solid lines are drawn to aid the eye. δ-Lysin concentration was 0.5 µM in all experiments.

LUVs made from bacterial lipid extracts behave markedly different compared to those based on the three types of synthetic phospholipids described above. PG extracts from S. aureus and S. sciuri produce lipid vesicles that retain encapsulated dye for days, similar to those prepared from POPG. δ-Lysin binds readily to PG vesicles and releases dye with kinetics that are generally slower than those observed in PC dispersions at the same lipid concentration (Figs. 3 and 5).

Figure 5.

Carboxyfluorescein efflux from POPG and bacterial lipid LUVs induced by δ-lysin. POPG, black line; S. aureus PG extract, red line; S. sciuri PG extract, blue line. Peptide concentration was 0.5 µM and lipid concentration, 200 µM.

PG extracts from S. sciuri are enriched in iso- over anteiso-branched fatty acids (Table 1). If the original hypothesis were correct and peptide-lipid interactions were indeed favored in bilayers enriched in iso-branched fatty acids, dye release should occur more rapidly from LUVs made from S. sciuri PG extracts than from those based on S. aureus extracts, at the same lipid concentration. However, within experimental error, dye release occurs with roughly the same kinetics from S. aureus, S. sciuri, and POPG lipid dispersions (Fig. 5).

Table 1.

Percent fatty acid composition of the major phosphatidylglycerol fractions isolated from S.aureus and S. sciuri membrane lipids. Palmitic, stearic, 17-methyl stearic, 19-methyl stearic, and arachidic acid make up the remaining fractions.

Fatty acid	S. aureus %	S. sciuri %
13-methyl myristic acid (i15:0)	4.5	43.7
12-methyl myristic acid (ai15:0)	47.4	30.7
15-methyl palmitic acid (i17:0)	5.4	10.9
14-methyl palmitic acid (ai17:0)	28.6	9.8

Only at lipid concentrations > 200µM do the kinetics of dye release from S. aureus lipid vesicles become significantly slower than from those based on S. sciuri lipid extracts (Fig. 6). We pointed out previously that the bacterial PG extracts still contained a significant fraction of lysine-modified PG (Lys-PG), which co-elutes with unmodified PG during purification under the chosen conditions (37). The presence of 10–20 mol% Lys-PG in mixtures with PG was found to lead to a stabilization of the membrane, rendering the membrane less susceptible to peptide-induced perturbations (37), presumably due to a lower headgroup repulsion in PG/Lys-PG mixtures compared with that in pure PG bilayers.

Figure 6.

Average kinetics (τ) of peptide-induced carboxyfluorescein efflux from POPG and bacterial lipid vesicles as a function of lipid concentration. POPG, black symbols; S. aureus PG extract, red symbols; S. sciuri PG extract, blue symbols. Solid lines are drawn to aid the eye. δ-Lysin concentration was 0.5 µM in all experiment.

Thus, the more complex mixture of lipids found in bacterial membranes appears to provide a hydrophobic barrier better suited to protect against the uncontrolled passage of small molecules. However, scanning electron microscopy images obtained from bacterial membranes that have been cooled to low temperatures support the notion that bilayers composed mostly of saturated branched lipids behave fundamentally differently from those composed of unsaturated lipids (36). When S. aureus whole membranes were cooled to below the phase transition (−10°C), no aggregation of proteins in the plane of the membrane was observed, whereas E. coli membranes, which contain unsaturated lipids, showed significant protein aggregation when cooled to low temperatures. This observation suggests that E. coli lipids form a well-packed gel-phase below the T_m from which proteins are excluded, in contrast to the S. aureus membranes, which show sufficient lateral compressibility even in the gel phase to prevent membrane protein aggregation.

We had originally hypothesized that peptide–lipid interactions would be disfavored in systems composed of lipids with low gel–fluid phase transition temperatures. In summary, we could show that asymmetric phospholipids containing iso-branched fatty acids indeed facilitate the formation of peptide-induced membrane defects when compared to bilayers comprised of the low-temperature melting anteiso-branched isomer, as we had hypothesized. We attribute this to the formation of peptide oligomers that is somewhat more favorable in iso-branched than in anteiso-branched lipids. The magnitude of the observed effect, however, depends strongly on the peptide-to-lipid ratio and becomes negligible at peptide-to-lipid ratio ≤ 1:400 (Fig. 6). Even at a peptide-to-lipid ratio of 1:600, the average τ of dye release is only increased by a factor of 5 in anteiso-relative to iso-branched lipids.

Thus, the magnitude of the effect is so small that the increase in the fraction of anteiso-branched fatty acids in antibiotic-resistant strains of S. aureus is unlikely to be responsible for the observed increased antibiotic resistance. Moreover, the correlation between T_m and ease of peptide-induced defect formation does not hold true in general. δ-Lysin-induced dye release from POPC vesicles (T_m = −3.5°C) is slower at every lipid concentration tested, than that from either 16:0-i15:0PC or 16:0-ai15:0PC vesicles (Fig. 4). Hence, the T_m of an aqueous lipid dispersion is a poor predictor of peptide activity in a lipid model system. However, there is a clear correlation between the overall integrity of the bilayer, as measured qualitatively by the ability to retain encapsulated dye, and its sensitivity to the action of a membrane-active peptide. This is especially evident in bilayers composed of 16:0-13:0PC that are unable to retain content for a period of time that is significant on a biological scale. Thus, bacterial membranes that rely on branched rather than unsaturated lipids to ensure a fluid-phase bilayer at physiological temperatures must include other membrane components to improve packing and integrity of the hydrophobic barrier.

Highlights.

• Saturated branched phospholipids of the type that occur in bacterial membranes have a main phase transition below room temperature.

• Lipid vesicles composed of branched phospholipids are significantly more susceptible to membrane-active peptides than those made from unsaturated lipids.

• Lipid vesicles composed of synthetic branched phospholipids do not retain encapsulated dye for extended periods of time.

• Both anteiso- and iso-branched phospholipids facilitate peptide-induced membrance perturbations compared to unsaturated phospholipids.

• Membrane phospholipids extracted from S.aureus give rise to liposomes that are more resistant than those made from pure synthetic lipids.